On the mechanism of the palladium-catalyzed β-arylation of ester enolates

Article abstract of DOI:10.1002/chem.201103153

The palladium-catalyzed β-arylation of ester enolates with aryl bromides was studied both experimentally and computationally. First, the effect of the ligand on the selectivity of the α/β-arylation reactions of ortho- and meta-fluorobromobenzene was described. Selective β-arylation was observed for the reaction of o-fluorobromobenzene with a range of biarylphosphine ligands, whereas α-arylation was predominantly observed with m-fluorobromobenzene for all ligands except DavePhos, which gave an approximate 1:1 mixture of α-/β-arylated products. Next, the effect of the substitution pattern of the aryl bromide reactant was studied with DavePhos as the ligand. We showed that electronic factors played a major role in the α/β-arylation selectivity, with electron-withdrawing substituents favoring β-arylation. Kinetic and deuterium-labeling experiments suggested that the rate-limiting step of β-arylation with DavePhos as the ligand was the palladium-enolate-to-homoenolate isomerization, which occurs by a β-H-elimination, olefin-rotation, and olefin-insertion sequence. A dimeric oxidative-addition complex, which was shown to be catalytically competent, was isolated and structurally characterized. A common mechanism for α- and β-arylation was described by DFT calculations. With DavePhos as the ligand, the pathway leading to β-arylation was kinetically favored over the pathway leading to α-arylation, with the palladium-enolate-to-homoenolate isomerization being the rate-limiting step of the β-arylation pathway and the transition state for olefin insertion its highest point. The nature of the rate-limiting step changed with PCy₃ and PtBu₃ ligands, and with the latter, α-arylation became kinetically favored. The trend in selectivity observed experimentally with differently substituted aryl bromides agreed well with that observed from the calculations. The presence of electron-withdrawing groups on these bromides mainly affected the α-arylation pathway by disfavoring C-C reductive elimination. The higher activity of the ligands of the biaryldialkylphosphine ligands compared to their corresponding trialkylphosphines could be attributed to stabilizing interactions between the biaryl backbone of the ligands and the metal center, thereby preventing deactivation of the β-arylation pathway.

Full text of DOI:10.1002/chem.201103153

DOI: 10.1002/chem.201103153
On the Mechanism of the Palladium-Catalyzed b-Arylation of Ester Enolates
Paolo Larini,[a] Christos E. Kefalidis,[b] Rodolphe Jazzar,[a] Alice Renaudat,[a]
Eric Clot,*[b] and Olivier Baudoin*[a]
Abstract: The palladium-catalyzed b-
arylation of ester enolates with aryl
bromides was studied both experimen-
tally and computationally. First, the
effect of the ligand on the selectivity of
the a/b-arylation reactions of ortho-
and meta-fluorobromobenzene was de-
scribed. Selective b-arylation was ob-
served for the reaction of o-fluorobro-
mobenzene with a range of biarylphos-
phine ligands, whereas a-arylation was
predominantly observed with m-fluoro-
bromobenzene for all ligands except
DavePhos, which gave an approximate
1:1 mixture of a-/b-arylated products.
Next, the effect of the substitution pat-
tern of the aryl bromide reactant was
studied with DavePhos as the ligand.
We showed that electronic factors
played a major role in the a/b-arylation
selectivity, with electron-withdrawing
substituents favoring b-arylation. Ki-
netic and deuterium-labeling experi-
ments suggested that the rate-limiting
step of b-arylation with DavePhos as
the ligand was the palladium–enolate-
to-homoenolate isomerization, which
pathway and the transition state for
olefin insertion its highest point. The
nature of the rate-limiting step changed
with PCy3 and PtBu3 ligands, and with
the latter, a-arylation became kineti-
cally favored. The trend in selectivity
observed experimentally with different-
ly substituted aryl bromides agreed
well with that observed from the calcu-
lations. The presence of electron-with-
drawing groups on these bromides
mainly affected the a-arylation path-
ꢀ
occurs by a b H-elimination, olefin-ro-
tation, and olefin-insertion sequence. A
dimeric oxidative-addition complex,
which was shown to be catalytically
competent, was isolated and structural-
ly characterized. A common mecha-
nism for a- and b-arylation was de-
scribed by DFT calculations. With Da-
vePhos as the ligand, the pathway lead-
ing to b-arylation was kinetically fa-
vored over the pathway leading to a-
arylation, with the palladium–enolate-
to-homoenolate isomerization being
the rate-limiting step of the b-arylation
ꢀ
way by disfavoring C C reductive
elimination. The higher activity of the
ligands of the biaryldialkylphosphine li-
gands compared to their corresponding
trialkylphosphines could be attributed
to stabilizing interactions between the
biaryl backbone of the ligands and the
metal center, thereby preventing deac-
tivation of the b-arylation pathway.
Keywords: arylation · density func-
tional calculations · isomerization ·
palladium · reaction mechanisms
Introduction
halide or pseudo-halide to an active palladium(0) catalyst,
followed by ligand substitution to give a palladium–enolate
intermediate, which provides the a-arylated product upon
reductive elimination (Scheme 1, path a). Isolation of palla-
dium-enolate intermediates and kinetic experiments have
been reported to support this catalytic cycle.[2] Site-selective
ꢀ
The palladium(0)-catalyzed C H arylation a to electron-
withdrawing groups such as carbonyls, nitriles, nitro groups,
and sulfones has been established as a powerful method for
the construction of Csp3 Csp2 bonds.[1] Mechanistically, this
ꢀ
ꢀ
reaction occurs through the oxidative addition of an aryl
direct arylation of Csp3 H bonds that are more remote from
[a] Dr. P. Larini,+ Dr. R. Jazzar,+ Dr. A. Renaudat, Prof. Dr. O. Baudoin
Universitꢀ Claude Bernard Lyon 1, CNRS UMR5246
Institut de Chimie et Biochimie Molꢀculaires
et Supramolꢀculaires
CPE Lyon, 43 Boulevard du 11 Novembre 1918
69622 Villeurbanne (France)
E-mail: olivier.baudoin@univ-lyon1.fr
[b] Dr. C. E. Kefalidis,+ Dr. E. Clot
Institut Charles Gerhardt, CNRS UMR5253
Universitꢀ Montpellier 2
case courrier 1501, Place Eugꢁne Bataillon
34095 Montpellier (France)
E-mail: clot@univ-montp2.fr
[+] These authors contributed equally to this work.
ꢀ
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201103153.
Scheme 1. Different mechanistic pathways leading to a- or b-C H aryla-
tion.
1932
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1932 – 1944

FULL PAPER
functional groups are highly desirable given the broad syn-
thetic utility of their corresponding products.[3] In this
ꢀ
regard, palladium-catalyzed C H arylations b to electron-
withdrawing functional groups have been reported, most of
which also employ aryl halides as convenient aryl-group
donors.[4] All of these methods seem to proceed via a C H
ꢀ
activation mechanism (Scheme 1, path c), wherein the func-
tional group serves as a directing group that enables the
direct formation of a palladium homoenolate, which gives
rise to the b-arylated product upon reductive elimination.
Building on a seminal example by Hartwig and co-work-
ers,[5] we recently reported our preliminary results on the b-
[6]
ꢀ
C H arylation of carboxylic esters. Our first experimental
and computational data indicated that this reaction occurs
through a pathway related to a-arylation (and therefore dis-
tinct to other b-arylation reactions) via the formation of an
intermediate palladium enolate, rearrangement to a palladi-
um homoenolate, and reductive elimination (Scheme 1,
path b). This alternative mechanism has both advantages
ꢀ
and drawbacks compared to the directed C H activation
mechanism. The advantages include the use of simple car-
boxylic esters as the reactants, which are precursors of a
wide range of other functional groups, milder reaction tem-
perature, and the fact that bis-arylation is not observed. The
drawbacks include restriction of the reaction scope to eno-
lizable esters, the need for a strong base to generate the
ester enolate, and competitive a-arylation as a side-reac-
tions. Herein, we report detailed mechanistic investigations,
including both experimental and theoretical data, which
Figure 1. Effect of the ligand on the conversion and on the a/b-selectivity
of the arylation of methyl isobutyrate (2a) with o-fluorobromobenzene
(1a).
ꢀ
shed light on this intriguing and synthetically useful C H
functionalization process and allows us to envisage new re-
action extensions.
Results and Discussion
Reaction optimization and a/b-arylation selectivity: The ary-
lation of the lithium enolate formed in situ from methyl iso-
butyrate (2a) and lithium dicyclohexylamide with 2-fluoro-
bromobenzene (1a) was investigated under conditions close
to those previously described by Hartwig and co-workers for
an a-arylation reaction.[5] The effect of the phosphine ligand
on a/b-arylation selectivity for a [Pd]/ligand ratio of 1:1 is
shown in Figure 1. With PtBu3 as the ligand, a-arylated
product 3a was obtained in an 85:15 ratio with compound
4a. This ratio was reversed with the less bulky tricyclohexyl-
phosphine ligand, which afforded complete conversion and
selectivity for the b-arylation product at 1108C. However,
attempts to decrease the reaction temperature failed with
this ligand. Next, we turned our attention to Buchwaldꢃs
biaryldicyclohexylphosphine ligands L1–L5 (Scheme 2).[7]
Apart from X-Phos (L3), all of these ligands furnished com-
plete conversion at 508C and high b/a selectivity. The high-
est selectivity was achieved with DavePhos (L2), which pro-
vided the product (4a) with no observable trace of com-
pound 3a; therefore, further optimization studies were con-
ducted with this ligand. Of the bases that were screened in
Scheme 2. Structures of biarylphosphine ligands used in this study.
combination with ester 2a, only the strongly basic lithium
amides (LDA, Cy2NLi, and LiTMP) gave complete reaction
conversion, which indicates that the lithium enolate of com-
pound 2a is indeed the reactive species. Lithium dicyclohex-
ylamide, which tended to give higher and more-reproducible
yields as reported previously,[5] was used for further studies.
Comparison of various solvents indicated that apolar,
non-coordinating solvents such as toluene furnished the
Chem. Eur. J. 2012, 18, 1932 – 1944
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1933

E. Clot, O. Baudoin et al.
highest yield of compound 4a. Different palladium precata-
lysts (e.g., [Pd2A(dba)3], Pd(OAc)2, and [{Pd(allyl)Cl}2]) were
evaluated in combination with different proportions of Da-
vePhos; a 1:2 ratio of [Pd2A(dba)3] (5 mol%)/DavePhos
have a methyl (L7), methoxy (L6), or a dimethylamino (L2)
substituent at the 2ꢃ position of the biphenyl scaffold, it ap-
pears that the more-electron-donating and/or better-coordi-
nating NMe2 group increases the selectivity towards b-aryla-
tion. On the other hand, if one compares mono-methoxy-
substituted ligand L6 with bis-alkoxy-substituted ligands
SPhos (L4) and RuPhos (L5), and mono-dimethylamino-sub-
stituted ligand DavePhos (L2) with bis-dimethylamino-sub-
stituted CPhos (L8),[10] it seems that the ligand flexibility at
the biaryl bond also plays an important role in the a/b-aryla-
tion selectivity, with decreased flexibility giving less b-arylat-
ed product. A further indication of the importance of flexi-
bility in the biaryl bond is shown by the trend in selectivity
observed with KenPhos (L10, Scheme 2),[11] a chiral bi-
naphthyl analogue of DavePhos for which biaryl-bond rota-
tion is blocked. Whereas ligand L10 furnished complete b-ar-
ylation selectivity with o-fluorobromobenzene (1a), together
with a promising 50% ee,[6] complete selectivity for the a-ar-
ylation product was again observed with m-fluorobromoben-
zene (1b, Figure 2). In addition, the impact of the other sub-
stituents on the phosphorus atom of DavePhos was briefly
considered, with ligands L11–L13 (Scheme 2, Figure 2).[12] In
the presence of the more-bulky tert-butyl groups instead of
the cyclohexyl groups (L11),[13–14] complete selectivity for the
a-arylation product was observed. On the other hand, with
less-bulky isopropyl groups (L12), the selectivity for the b-ar-
ylation product decreased (56:44 ratio in favor of compound
3b) compared to that with ligand L2. When the both less-
bulky and electron-rich analogue L13 was used, which con-
tained two phenyl substituents, complete selectivity for the
a-arylation product was again observed. These results show
that DavePhos (L2) possesses the optimal electronic and
steric properties among the biarylphosphine ligands that
were evaluated in this study of the b-arylation of methyl iso-
butyrate (2a) with m-fluorobromobenzene (1b). Finally,
modifying the reaction conditions with the [Pd]/DavePhos
catalytic system did not seem to affect the a/b selectivity. In
particular, changing the solvent or varying the reaction tem-
perature from 30–1108C did not modify the a/b ratio to an
appreciable extent. We are currently exploring further struc-
tural modifications of DavePhos to better understand these
electronic and conformational effects and to improve the b-
arylation selectivity.
G
E
ACHTUNGTRENNUNG
CTHUNGTRENNUNG
(10 mol%) provided the highest yields with a range of sub-
strates. A final adjustment of the ratio of enolate to aryl
bromide provided the following optimal conditions, which
afforded the isolated product 4a in 69% yield: ester 2a
(1.6 equiv), Cy2NLi (1.7 equiv), aryl bromide 1a (1 equiv),
[Pd
308C.[6]
Next, we studied the effect of the position of the fluorine
2ACHTUNGTRENNUNG
(dba)3] (5 mol%), DavePhos (L2, 10 mol%), toluene,
atom with respect to the bromine atom on the aryl bromide.
With the [Pd]/DavePhos catalyst, moving the fluorine sub-
stituent from the ortho- to the meta- or para position dra-
matically affected the a/b-arylation selectivity, from >99%
b-arylation with an ortho-fluorine atom (Figure 1) to 47:53
a/b-arylation with
a
meta- or para-fluorine atom
(Figure 2).[6] We decided to re-examine the effect of the
ligand in the reaction of m-fluorobromobenzene (1b) with
methyl isobutyrate (2a, Figure 2).
Surprisingly, from a range of commercially available or
readily accessible[8] biaryldicyclohexylphosphines (L1–L9),
only DavePhos (L2) and its methoxy analogue (L6)[9] provid-
ed a significant amount of b-arylated product 4b (53% and
26%, respectively), whereas all other ligands gave complete
selectivity in favor of a-arylated product 3b. This observa-
tion stands in sharp contrast to that made with o-fluorobro-
mobenzene (Figure 1), for which various biaryldicyclohexyl-
phosphine ligands gave a high selectivity for b-arylation. By
comparing unsubstituted Cy-JohnPhos (L1) with ligands that
With these data in hand, we studied the trend in selectivi-
ty for other aryl bromides at temperatures in the range 30–
508C (Table 1).[15] For this study, tert-butyl isobutyrate (2b)
was chosen instead of methyl isobutyrate (2a) because the
former gave cleaner reaction mixtures because of the sup-
pression of subsequent Claisen-condensation reactions of
the reaction products.[5] We confirmed that compounds 2a
and 2b gave comparable a/b-arylation selectivities over a
representative range of aryl bromides. In our first communi-
cation,[6] we showed that only ortho-electronegative substitu-
ents on the aryl bromide, such as F, Cl, CF3, OCF3, and
OMe groups provided complete selectivity for b-arylation in
the reaction with esters 2a or 2b. Table 1 shows additional
examples that allow for a deeper analysis of the impact of
Figure 2. Effect of the ligand on the conversion and on the a/b selectivity
of the arylation of methyl isobutyrate (2a) with m-fluorobromobenzene
(1b).
1934
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1932 – 1944

Mechanism of the Palladium-Catalyzed b-Arylation of Ester Enolates
FULL PAPER
Table 1. Effect of the aryl-bromide structure on the b/a selectivity for the arylation of tert-butyl isobutyrate
(2b).
the substitution pattern on the
aryl-bromide on the a/b-aryla-
tion selectivity. Table 1, en-
tries 1–3 show the effect of the
position of
a fluorine atom
with respect to the bromine
atom on the a/b-arylation se-
lectivity (cf. Figure 1, Figure 2).
With bromobenzene (1d), a-
arylated product 3d was the
major arylation product and
was obtained in an 82:18 ratio
(Table 1, entry 4). As with bro-
Entry
1
Aryl bromide
Products
a/b ratio[a]
<2:98
Yield [%][b]
63[c]
mobenzene,
o-bromotoluene
2
47:53
95
(1e) and m-bromotoluene (1 f)
mainly afforded their corre-
sponding a-arylated products
(3e and 3 f, respectively) albeit
in low yields (Table 1, entries 5
and 6), which, in comparison
to compounds 1a and 1b
(Table 1, entries 1 and 2), indi-
cates that electronic effects
mainly govern the a/b-aryla-
tion selectivity with respect to
the aryl bromide substrate. As
shown by the comparison of
Table 1, entries 5 and 6, steric
effects have a minor but signif-
icant impact on the selectivity,
with an ortho substituent favor-
ing b-arylation. Comparison of
the reactions of ortho-, meta-,
and para-methoxybromoben-
zene (1g–1i; Table 1, entries 7–
9) shows that when the me-
thoxy group was in the ortho
position, a high selectivity for
b-arylation was observed, simi-
lar to the ortho-fluorine group,
and almost no effect when the
methoxy group was in the
meta- and para positions (cf.
Table 1, entry 4). The ortho
effect of the OMe group may
be ascribed to its electron-
withdrawing (inductive) char-
acter at this position, whereas
its electron-donating (reso-
nance) character at the meta-
and para positions does not
seem to have an influence on
the selectivity. The reactions of
mixed fluoro/methoxy sub-
strates 1j and 1k (Table 1, en-
tries 10 and 11, respectively)
showed the same trend in se-
3
4
47:53
82:18
91
83
5
6
74:26
89:11
15
30
7
8
<1:99
64[c]
84:16
35
9
80:20
40
10
<1:99
74[c]
11
47:53
76
12
13
40:60
63:37
98
75
Chem. Eur. J. 2012, 18, 1932 – 1944
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1935

E. Clot, O. Baudoin et al.
terized by single-crystal X-ray
diffraction (Figure 3a).[19] X-
ray structures of oxidative-ad-
dition [Pd] complexes with
biarylphosphine ligands are
scarce,[20] and to the best of our
knowledge none has been re-
ported with DavePhos as the
ligand. Complex IIb shows a m-
bromo dimeric structure that
features a very weak interac-
tion between the palladium
Table 1. (Continued)
Entry
Aryl bromide
Products
a/b ratio[a]
Yield [%][b]
14
9:91
85
15
5:95
57
1
1
[a] Measured by H or 19F NMR spectroscopy, or GCMS. [b] Measured by H NMR spectroscopy by using tri-
chloroethylene as an internal standard. [c] Yield of the isolated b-arylated product.
atom and the apical nitrogen
[21]
ꢀ
atom (Pd N distance 3.2 ꢄ).
lectivity as those without a methoxy substituent (Table 1, en-
tries 1 and 2); this result further confirms the lack of effect
of the electron-donating para-methoxy group on the selec-
tivity of the reaction.
Next, we examined the effect of a trifluoromethyl group
at the meta and para positions was examined (Table 1, en-
tries 12 and 13). As expected from stronger inductive effects,
higher selectivity for the b-arylation product was observed
with m-CF3-substituted 1l. In line with this result, and owing
to the cumulative nature of inductive effects, aryl bromide
1n , which contained two meta-CF3 groups, furnished a
much-higher b/a-arylation ratio (Table 1, entry 14). A simi-
lar trend was observed with 3,4,5-trifluorobromobenzene
(1o, Table 1, entry 15), which provided a 95:5 ratio in favor
of b-arylated product 4o. These above results demonstrate
that the effect of the substituents on the aryl bromide on
the selectivity of the a/b-arylation reaction is mainly elec-
tronic in nature, with electron-withdrawing substituents
giving more b-arylated product and no substituent or elec-
tron-donating substituents giving more a-arylated product.
For a given electron-withdrawing substituent such as CF3,
Scheme 3. Mechanism of the a- and b-arylation of methyl isobutyrate 2a
with the Pd/DavePhos catalyst (Ar=2-fluorophenyl).
this effect is maximized when the substituent is at the ortho
position,[6] and decreases when moving this group to meta
and para positions. These electronic effects are in line with
DFT calculations and will be further commented on below.
Comparison of the kinetics of the reactions of compounds
1a and 2a, catalyzed by IIb (10 mol%) or [Pd
Phos (5 mol%/10 mol%), showed that complex IIb is about
three times faster than the [Pd2A(dba)3]/DavePhos combina-
2ACHTUNGTRENUN(NG dba)3]/Dave-
Experimental mechanistic studies: We have previously pro-
posed a common mechanism for a- and b-arylation reactions
based on experimental observations and DFT calculations
for the reaction of methyl isobutyrate (2a) and o-fluorobro-
CTHUNGTRENNUNG
tion; thus, complex IIb is a competent catalyst for this reac-
tion (see the Supporting Information, Figure S1).[19] Dimer
IIb is likely in fast equilibrium with monomeric species IIa,
which lies within the catalytic cycle. A kinetic experiment
involving increasing concentrations of dba showed that the
dba has an inhibitory effect on the reaction (see the Sup-
porting Information, Figure S2). This known inhibitory
effect[23] may be at least partly responsible for the higher re-
activity observed with isolated complex IIb compared to the
mobenzene (1a) using the [Pd]/PCy3 catalytic system.[6]
A
modified mechanism for the same reaction catalyzed by
[Pd]/DavePhos is depicted in Scheme 3.
It is well-known that the active catalyst formed with bulky
biarylphosphine ligands, such as L2, in cross-coupling reac-
tions is a monoligated palladium(0) complex (Ia), which is
in equilibrium with the less-reactive [Pd0(L)2] species
(Ib).[7,16,17] In an elementary first step, oxidative addition of
aryl bromide 1a to complex Ia generates complex IIa. Oxi-
dative-addition complex IIb was prepared from a mixture of
[Pd
ligand instead of DavePhos, monometallic complex
[Pd(Ar)Br(PCy3)2] (IIc) was isolated from the oxidative ad-
dition of compound 1a to [Pd(PCy3)2] (Figure 3b). Howev-
er, complex IIc was not a competent catalyst for the b-aryla-
2ACHTUNGERTN(NUNG dba)3]/DavePhos in situ mixture. With PCy3 as the
AHCTUNGTRENNUNG
[Pd
G
G
ACHTUNGTRENNUNG
Phos (L2), and o-fluorobromobenzene (1a) and was charac-
1936
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1932 – 1944

Mechanism of the Palladium-Catalyzed b-Arylation of Ester Enolates
FULL PAPER
Figure 3. X-ray crystal structures of the oxidative-addition palladium-
complexes a) IIb, obtained with compound 1a and DavePhos, and b) IIc,
obtained with compound 1a and PCy3 (thermal ellipsoids set at 30%
probability, H atoms were omitted for clarity).[22]
Figure 4. Logarithmic plot of the initial rate dependence on the concen-
tration of a) o-fluorobromobenzene (1a), b) the lithium enolate of 2a
(generated in situ from a 1:1 ratio of compound 2a and LiNCy2).[19]
ment with added Cy2NH showed that the free amine indeed
contributes significantly to the observed negative order in
enolate, although this inhibition is less pronounced than that
induced by the lithium enolate itself (see the Supporting In-
formation, Figure S3).[19] This reaction inhibition observed
with the enolate of compound 2a, Cy2NH, and dba might in-
dicate that there are several catalyst-deactivation pathways.
In this regard, more-precise information would be obtained
by performing detailed kinetic and spectroscopic experi-
ments; these experiments will be the subject of our future
investigations. After ligand substitution (Scheme 3), palladi-
um enolate III may follow two pathways: 1) reductive elimi-
tion reaction of compound 2a with compound 1a, even at
elevated temperatures, presumably because it was unable to
generate the active [Pd(Ar)(Br)L] monoligated species,
unlike complex IIb. This result may explain why the [Pd2-
ACHTUNGTRENNUNG(dba)3]/PCy3 combination is only a competent catalyst for b-
arylation at high temperatures whereas the [Pd]/DavePhos
system can operate at room temperature (Figure 1).
The reaction order in o-fluorobromobenzene (1a) was de-
termined by plotting the log of the initial rate versus the log
of the concentration of compound 1a (Figure 4a). A slope
of 0.08 was obtained, thereby indicating that the reaction is
zero-order in compound 1a; therefore, oxidative addition is
not rate-limiting. This observation is to be expected from
the electronic properties of both the ligand (electron-rich)
and the aryl bromide (electron-poor).[7]
The next step of the catalytic cycle is the substitution of
the bromine atom with the lithium enolate generated in situ
from ester 2a and LiNCy2 (Scheme 3). A negative order was
established for the lithium enolate (Figure 4b), which shows
that the reaction is inhibited by increasing the concentration
of enolate and thus that the ligand-substitution step is not
rate-determining either. Dicyclohexylamine is produced in
stoichiometric amounts with the lithium enolate of com-
pound 2a; therefore, dicyclohexylamine could be also re-
sponsible for the observed reaction inhibition. An experi-
ꢀ
nation to give a-arylated product 3a, or 2) b H elimination
to give olefin complex IVa.[24] Bond rotation provides iso-
meric complex IVb, which can undergo olefin insertion
ꢀ
ꢀ
either into the Pd H or the Pd Ar bonds. Previous calcula-
tions with PCy3 as the ligand indicated that the former inser-
tion is much-more-kinetically favored than the latter.[6] The
corresponding palladium homoenolate V then undergoes re-
ductive elimination to give b-arylated product 4a and initial
complex Ia.
To gain an insight into the steps following ligand substitu-
tion (i.e., isomerization of palladium enolate III into homoe-
nolate V and reductive elimination; Scheme 3), the reactivi-
ty of deuterium-labeled esters 2c–2e was analyzed
(Scheme 4).[25] The reaction of ester 2c, which had a deuteri-
Chem. Eur. J. 2012, 18, 1932 – 1944
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1937

E. Clot, O. Baudoin et al.
overall isomerization of palladium enolate III into palladi-
ꢀ
um homoenolate V, involving the b H-elimination/olefin-ro-
tation/olefin-insertion sequence,[27] is rate-determining.[28] At
this stage, DFT calculations were deemed necessary to com-
plete this mechanistic picture. In particular, these calcula-
tions were expected to provide more insight into the relative
influence of the different elementary steps on the reaction
rate and on the selectivity for a-/b-arylation.
Computational studies: The mechanism of the reaction of
compound 1a with compound 2a, catalyzed by [Pd-
AHCTUNGTREG(NNNU DavePhos)], was studied computationally at the B3PW91
level.[29] The actual biarylphosphine ligand DavePhos was
considered in the calculations. The extrema were located in
the gas phase and the energies given below are polarizable
continuum model (PCM) electronic energy values (toluene)
corrected by gas phase Gibbs free energy contributions.[19]
ꢀ
As the C Br oxidative-addition and ligand-substitution
steps were shown experimentally not to be rate-determining,
the computational study concentrated on the sequence of
steps from enolate complex III to the regeneration of active
catalyst Ia (Scheme 3). Given the asymmetry of the Dave-
Phos ligand and of the ortho-fluoro-substituted aryl group of
compound 1a, there are in principle eight different possible
geometries for enolate III. For the optimization of six of
these possibilities, we assumed that the situation where the
ester OMe group points toward the biaryl moiety was disfa-
vored on steric grounds (see the Supporting Information,
Figure S4).[19] In most cases, the energetically preferred geo-
metries were where the ortho-fluoro group was on the same
side of the molecule as the biaryl group of DavePhos. In ad-
dition, the ester OMe group preferred a position that avoid-
ed the biaryl group on the phosphine ligand (IIIa; Figure 5).
The mechanisms of the a- and b-arylation reactions were
computed for the most-stable isomer (IIIa). The geometry
of isomer IIIa is pseudo-square-planar with the enolate
Scheme 4. b-Arylation experiments with deuterated esters. Reaction con-
ditions: ester (1.6 equiv), Cy2NLi (1.7 equiv), 2-chlorobromobenzene (1p,
1.0 equiv), [Pd2ACHTUNGTRENNUNG
(dba)3] (5 mol%), DavePhos (L2, 10 mol%), toluene,
258C. Kinetic isotope effects were measured by 1H NMR spectroscopy.
Cy=cyclohexyl, dba=dibenzylideneacetone.
um atom at the a-position, gave product 4p, which showed
no trace of deuterium (Scheme 4a). From starting ester 2d,
which had fully deuterated b-positions and a protonated a-
position, product 4q showed a complete D shift onto the a-
position (Scheme 4b). These two experiments are consistent
with the proposed mechanism (Scheme 3) and rule out a di-
ꢀ
rected-C H-activation mechanism (Scheme 1, path c). An
intramolecular kinetic isotope effect (KIE) was then deter-
mined from the reaction of ester 2e, which provided a kH/kD
value of 4.0 (Scheme 4c).[26] The corresponding intermolecu-
lar KIE value was measured by reacting a 1:1 ratio of esters
2 f and 2d and analyzing the ratio of products 4p and 4q at
various conversions (Scheme 4d). A constant 4p/4q ratio of
1.2 was obtained, which corresponded to a kH/kD value of
1.2. The important difference between the intra- and inter-
ꢀ
molecular isotope effects indicates that the b H-elimination
(III!IVa) or olefin-insertion steps (IVb!V) alone are not
rate-determining. However, the significant intermolecular
ꢀ
effect (kH/kD >1) implies that either the b H-elimination or
the olefin-insertion steps has a significant impact on the re-
action rate.
Overall, the kinetic experiments indicate that the first ele-
mentary steps of the catalytic cycle (Scheme 3), that is oxi-
dative addition and ligand substitution, are not rate-deter-
mining. The impact of the subsequent steps on the reaction
kinetics, as assessed by deuterium-labeling experiments,
seems ambiguous at the moment. The magnitude of the
intra- and intermolecular isotope effects suggests that the
Figure 5. Optimized geometry and Gibbs free energy (kcalmolꢀ1, relative
to complex IIIa) of the various intermediates and transitions states along
the a-arylation pathway.
1938
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1932 – 1944

Mechanism of the Palladium-Catalyzed b-Arylation of Ester Enolates
group bonded through the a-carbon atom trans to the phos-
FULL PAPER
ꢀ
ꢀ
ꢀ
phorous atom (Pd Ca 2.082 ꢄ, Pd Ar 2.016 ꢄ); a b CH
ꢀ
agostic interaction (Cb H 1.160 ꢄ, H···Pd 1.909 ꢄ) com-
pletes the coordination sphere around the PdII center. The
pathway for a-arylation starts with an isomerization from
compound IIIa into isomer IIIb, which is an intermediate
that features a Pd···O interaction trans to the aryl group
(2.281 ꢄ). This intermediate is 2.1 kcalmolꢀ1 higher in
energy than isomer IIIa; the transition state (TS; TS-IIIa-
IIIb) is associated to rotation of the enolate ligand around
ꢀ1
ꢀ
the Pd Ca bond and is 4.7 kcalmol higher in energy than
ꢀ
isomer IIIa. From isomer IIIb, the C C reductive coupling
that leads to a-arylated product 3a proceeds through TS-
IIIb-a with a Gibbs free energy that is 23.5 kcalmolꢀ1 higher
than isomer IIIa (Figure 5). In transition state TS-IIIb-a, the
ꢀ
ꢀ
Pd Ar bond has elongated to 2.060 ꢄ, whilst the Pd Ca dis-
tance is now 2.314 ꢄ. The Pd···O interaction is lost in the TS
ꢀ
and the Ca CAr bond that forms is long (2.006 ꢄ). From
isomer IIIa, the reductive elimination is exoergic, with a
Gibbs free reaction energy DG=ꢀ24.9 kcalmolꢀ1 in favor
of a-arylated product 3a.
The pathway leading to b-arylated product 4a involves
several steps. The agostic enolate complex IIIa leads to an
ꢀ
olefin hydride intermediate (IVa) through a b H-elimina-
tion TS (TS-IIIa-IVa); this intermediate is 10.3 kcalmolꢀ1
higher in energy than complex IIIa (Figure 6) and is barely
more stable than the preceding transition state TS-IIIa-IVa,
which is 11.0 kcalmolꢀ1 higher in energy than complex IIIa.
ꢀ
The TS for b H elimination is late, as confirmed by the geo-
metrical parameters (Cb···H 1.946 ꢄ and H···Pd 1.586 ꢄ),
ꢀ
and the Ca Cb bond has shortened from 1.503 ꢄ in complex
IIIa to 1.409 ꢄ in TS-IIIa-IVa. The geometrical parameters
for intermediate IVa are only slightly different from those
ꢀ
ꢀ
for TS-IIIa-IVa (Cb···H 1.997 ꢄ, H Pd 1.585 ꢄ, Ca Cb
1.406 ꢄ). From intermediate IVa, olefin rotation leads to in-
termediate IVb through transition state TS-IVa-IVb
(Figure 6). The activation barrier for the rotation is 7.5 kcal
molꢀ1 and intermediate IVb is 6.8 kcalmolꢀ1 less stable than
intermediate IVa. In transition state TS-IVa-IVb, the olefin
Figure 6. Optimized geometry and Gibbs free energy (kcalmolꢀ1, relative
to IIIa) of the various intermediates and TSs along the b-arylation path-
way.
ꢀ
is roughly perpendicular to the Pd Ar axis (Ar-Pd-Cb-Ca
748). Once the olefin has experienced a 1808 rotation, inser-
ꢀ
ꢀ
ꢀ
tion into the Pd H bond through TS-IVb-V affords homoe-
with a Pd Cipso distance of 3.448 ꢄ. Whereas the C C cou-
pling step from compound IIIb to give a-arylated product
3a has an activation barrier of DG# =21.4 kcalmolꢀ1
nolate intermediate V in which the Cb atom is now attached
to the Pd center. This insertion is easy (DG# =2.7 kcalmolꢀ1)
and exoergic (DG=ꢀ13.1 kcalmolꢀ1). Homoenolate V is
only 4.0 kcalmolꢀ1 less stable than enolate IIIa despite the
fact that no stabilizing agostic interaction is present in com-
pound V (Figure 6). One important feature of homoenolate
V is the significantly stronger interactions between the
biaryl group of DavePhos and the Pd center, as illustrated
ꢀ
(Figure 5), the TS for C C coupling that leads to b-arylated
product 4a (TS-V-b) is only 9.0 kcalmolꢀ1 higher in energy
than compound V (Figure 6). This lower activation barrier
ꢀ
for C C coupling can be explained by two factors: 1) the
Pd···Cipso distance in transition state TS-V-b (3.178 ꢄ) sug-
gests a stabilization of the developing Pd0 character by the
biaryl group of DavePhos;[17,30] 2) there is also an a-agostic
by the much-shortened Pd···Cipso distance of 2.672 ꢄ (Cipso
=
ꢀ
ipso-carbon atom of the phenyl ring of DavePhos that bears
interaction (Cb H 1.116 ꢄ, H···Pd 2.102 ꢄ) that contributes
ꢀ
an NMe2 substituent). The Pd Cipso distances are about
to stabilize transition state TS-V-b (Figure 6). The reductive-
elimination step, V!Ia+4a, is exoergic, with a Gibbs free
energy of ꢀ33.7 kcalmolꢀ1. With respect to complex IIIa, b-
arylated product 4a lies at ꢀ29.7 kcalmolꢀ1 and is thus the
thermodynamic product of the reaction.
3.6 ꢄ for all of the extrema along the a-arylation pathway,
and the same value is obtained for the extrema preceding
compound V in the b-arylation pathway. Only transition
state TS-IVb-V starts to show some increased interaction,
Chem. Eur. J. 2012, 18, 1932 – 1944
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1939

E. Clot, O. Baudoin et al.
Cy could be ascribed to the
smaller steric impact of PCy3
compared to DavePhos. This
observation is further demon-
strated by the relative energy
of isomer IIIb-Cy, which is
1.1 kcalmolꢀ1 lower in energy
than isomer IIIa-Cy. The
Pd···O distance is shorter in
isomer IIIb-Cy (2.246 ꢄ) than
in compound IIIb (2.281 ꢄ).
The a-arylation pathway is
barely affected, as transition
state TS-IIIb-a-Cy is 23.8 kcal
molꢀ1 higher in energy than
IIIa-Cy. The thermodynamic
driving force is not as strong as
when DavePhos was used as
the ligand; the Gibbs free reac-
tion energy to form [PdACHTUNGTRENNUNG(PCy3)]
and 3a is DG=ꢀ13.5 kcal
molꢀ1.
value
The
more-negative
Figure 7. Comparison of the energy profiles for the a- and b-arylation pathways from enolate complex IIIa
with DavePhos as the ligand.
(DDGꢁ10 kcalmolꢀ1)
when DavePhos is used as the
A comparison of the a- and b-arylation pathways is
shown in Figure 7. In addition to the thermodynamic prefer-
ence for compound 4a over compound 3a, there is a clear
kinetic preference for the b-arylation pathway (DDG# =
3.7 kcalmolꢀ1). These calculated profiles are in perfect
agreement with the experimental observations, where only
compound 4a is observed in the reaction between com-
ligand (DG=ꢀ24.9 kcalmolꢀ1) is due to an extra stabiliza-
tion of the [PdACTHUNTGRNEUNG(DavePhos)] complex through coordination
of the aromatic ring that is not bonded to the phosphorous
atom.[30] The b-arylation pathway with [Pd
AHCTUNGTRENNUNG
decreased steric impact of the PCy3 ligand compared to Da-
vePhos; Figure 8 shows a comparison of the pathways with
each phosphine. The b H-elimination and olefin-rotation
ꢀ
pounds 1a and 2a, catalyzed by [Pd
N
steps are slightly easier with PCy3 as the ligand. The most-
significant difference occurs for the rotated olefin intermedi-
ates IVb and IVb-Cy: In intermediate IVb, the COOMe
ester group is proximal to the biphenyl group of DavePhos
and, consequently, the intermediate is 17.1 kcalmolꢀ1 higher
in energy than compound IIIa. Such steric repulsion is not
so stringent in intermediate IVb-Cy where the Cy group is
smaller and the energy of intermediate IVb-Cy relative to
IIIa-Cy is 12.5 kcalmolꢀ1. This difference is shown in the
arylation pathway, the transition states TS-IIIa-IVa and TS-
IVa-IVb lie at almost the same energies as their correspond-
ꢀ
ing products (IVa and IVb, respectively). Thus, the b H-
elimination, olefin-rotation, and olefin-insertion steps ac-
tually behave as a single step that corresponds to the iso-
merization of Pd enolate IIIa into homoenolate V. This iso-
merization is the rate-limiting step of the b-arylation path-
way, with an activation barrier of DG# =19.8 kcalmolꢀ1 (cal-
culated as the energy difference between compound IIIa
and the transition state TS-IVb-V). This analysis is in agree-
ment with the intra- and intermolecular kinetic isotope ef-
fects observed experimentally (Scheme 4c,d). The 19.8 kcal
molꢀ1 activation barrier is also in agreement with the reac-
tion occurring at room temperature.
ꢀ
TSs for insertion into the Pd H bond, where TS-IVb-V is
19.8 kcalmolꢀ1 higher in energy than compound IIIa where-
as TS-IVb-V-Cy is only 13.8 kcalmolꢀ1 higher in energy than
compound IIIa-Cy. In the case of DavePhos, the insertion
TS is the highest point along the b-arylation pathway,
whereas with PCy3 the TS for the olefin rotation is the high-
est point. However, in the latter case there is no stabilizing
influence of the phosphine ligand, neither in the homoeno-
To study the influence of the biaryl group in DavePhos on
the selectivity, the pathways for the a- and b-arylation reac-
ꢀ
tions were computed for [Pd
enolate complex IIIa-Cy presents the same overall geometry
as that obtained for compound IIIa. There is a clear b CH
G
late intermediate V-Cy nor in the TS for C C reductive
elimination (TS-V-b-Cy). Homoenolate V-Cy is 5.9 kcal
molꢀ1 less stable than enolate IIIa-Cy, whereas enolate V is
only 4 kcalmolꢀ1 less stable than compound IIIa. This differ-
ꢀ
agostic interaction (Cb H 1.166 ꢄ, H···Pd 1.896 ꢄ), and the
ꢀ
bonds to Pd are similar to those obtained in compound IIIa
ence in stability is enhanced in the TS of the C C-coupling
ꢀ
ꢀ
(Pd Ca 2.088 ꢄ, Pd Ar 2.010 ꢄ). The only significant differ-
step, where effective interaction with DavePhos lays TS-V-b
at 13.0 kcalmolꢀ1 higher in energy than compound IIIa,
whilst TS-V-b-Cy is 16.1 kcalmolꢀ1 higher in energy than
IIIa-Cy. With PCy3, the TS in the reductive-elimination step
ꢀ
ence is in a shorter Pd P bond in compound IIIa-Cy (Pd P
2.405 ꢄ for compound IIIa versus Pd P 2.376 ꢄ for com-
pound IIIa-Cy). The shorter Pd P bond in compound IIIa-
1940
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1932 – 1944

Mechanism of the Palladium-Catalyzed b-Arylation of Ester Enolates
FULL PAPER
arylation pathway and now the
selectivity has been reversed
with a DDG# value of 0.2 kcal
molꢀ1 in favor of the formation
of a-arylated product 3a. How-
ever, this value (0.2 kcalmolꢀ1)
does not correspond to the
85:15 ratio observed experi-
mentally, but the qualitative
trend is very well reproduced
by the calculations.
The reaction profiles with
[Pd
(DavePhos)]
and
[Pd-
AHCTUNRTGEG(NUNN PCy3)] (Figure 8) do not ex-
plain why the reaction with
PCy3 is not efficient at 508C
and why heating to 1108C is
necessary. In palladium-cata-
lyzed cross-coupling reactions,
the efficiency of the catalysis is
related to the ease with which
the necessary reactive species
Figure 8. Comparison of the energy profiles for the b-arylation pathway from enolate complex IIIa with three
different phosphine ligands.
ꢀ
are formed. For the C X oxi-
dative-addition step, a monoli-
is only 0.9 kcalmolꢀ1 below the TS of the olefin rotation and
the nature of the rate-determining step is different com-
pared to DavePhos. However, the b-arylation pathway is
also clearly preferred both kinetically and thermodynamical-
ly over the a-arylation pathway (DDG# =6.8 kcalmolꢀ1 and
DDG=4.3 kcalmolꢀ1, respectively), which is in agreement
with the experimental observation that only compound 4a is
produced in the reaction between 1a and 2a, catalyzed by
gated [Pd(L)] complex is preferred, hence bulky phosphines,
like PtBu3, are efficient ligands for this step.[33] Once the oxi-
dative addition has been performed, it is necessary to main-
tain a vacant site on the [Pd(Ar)(L)(X)] complex for substi-
tution of X by the other partner, in this case the enolate,
along an associative pathway.[34] This vacant site may be
trapped by a coordinating solvent, a free phosphine, or by
another molecule of [Pd(Ar)(L)(X)], thus forming a dimer
as observed experimentally with DavePhos (Figure 3). This
dimer has been optimized and the computed geometry of
compound IId is in very good agreement with the experi-
mental structure IIb (Figure 9). The computed N···Pd dis-
tance is longer than the experimentally determined value
(3.34 ꢄ versus 3.2 ꢄ). Dissociation of the dimer, whilst
maintaining the vacant site trans to the Ar group affords
compound IIe that is 7.1 kcalmolꢀ1 higher in energy than
compound IId. The N···Pd distance is shortened to 3.07 ꢄ in
compound IIe. However, as already shown by Barder et al.
for other biarylphosphines,[17] the DavePhos ligand is able to
form a secondary interaction between the biaryl group and
the metal. This interaction is indeed effective in compound
[PdACHTUNGTRENNUNG(PCy3)].
Thus, the b-arylation pathway seems to be strongly affect-
ed by the steric bulk of the phosphine ligand, whereas the
a-arylation pathway seems to be less sensitive to the steric
demands of the ligand. Therefore, if drastic modifications
are made on the phosphine ligand, an inversion of the selec-
tivity may be obtained. This is indeed what is observed ex-
perimentally in the reaction of compounds 1a and 2a cata-
lyzed by PtBu3 (Figure 1), where a-arylated product 3a is
obtained preferentially over b-arylated product 4a (85:15
ratio). The a- and b-arylation pathways were computed with
the PtBu3 ligand and the first significant change observed is
the easier overall a-arylation reaction pathway, with transi-
tion state TS-IIIb-a-tBu lying at 21.2 kcalmolꢀ1 higher in
energy than compound IIIa-tBu. This result could be ex-
plained by a larger steric impact of this phosphine ligand,
which facilitates the reductive-elimination step.[1a,31] The
easier reductive-elimination step with the bulkier PtBu3
ligand is also observed in the b-arylation pathway, where
TS-V-b-tBu lies at 13.9 kcalmolꢀ1 higher in energy than
compound IIIa-tBu (Figure 8). However, the steric impact
of PtBu3 is clearly evident on the TS for olefin rotation:
transition state TS-IVa-IVb-tBu is 21.4 kcalmolꢀ1 higher in
energy than compound IIIa-tBu, that is 4 kcalmolꢀ1 higher
than with PCy3.[32] This TS is the highest point along the b-
Figure 9. Optimized geometry for the dimer [{Pd
(DavePhos)}2] (IId) and the two isomers for the monomer [Pd
FC6H4)(Br)(DavePhos)] (IIe and IIf).
ACHUTGTNERN(NUG 2-FC6H4)ACHTUNGTRENNUNG
G
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
Chem. Eur. J. 2012, 18, 1932 – 1944
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1941

E. Clot, O. Baudoin et al.
ꢀ
IIf (Figure 9), an isomer of compound IIe, where a close
contact is evident between the unsubstituted ortho carbon
and Pd atoms (Pd···Cortho 2.572 ꢄ). As a result, compound
IIf is 12.3 kcalmolꢀ1 more stable than compound IIe and the
equilibrium IIdQIIf+IIf is strongly exoergic (DG=
ꢀ17.5 kcalmolꢀ1). The biaryl···Pd interaction effectively
shields the metal center from adventitious coordination that
would reduce the catalytic activity. The exoergicity of the
equilibrium for dimer dissociation is certainly overestimated
as the entropy contributions in the condensed phase are
lower than the values calculated in the gas phase. Neverthe-
less, the PCM value for the equilibrium (DE=0.2 kcal
molꢀ1) clearly indicates that the dimer easily dissociates.
This observation is in agreement with the fact that dimer IIb
is a competent catalyst for the b-arylation reaction of eno-
lates.
being the TS for olefin insertion into the Pd H bond. Con-
sequently, for the other aryl bromides, only the insertion TS
was computed together with the Pd–enolate complex and
the TS for the reductive-elimination step along the a-aryla-
tion pathway. Table 2 shows the activation barriers for the
a- and b-arylation pathways, and displays the computed and
experimental DDG# =DG# ꢀDG# values.
b
a
Table 2. Computed activation barriers (kcalmolꢀ1) along the a- (DG#
)
a
and b-arylation (DG#b) pathways, and comparison of calculated and ex-
perimental kinetic selectivities for the a- and b-arylation reactions
(DDG#).
[a]
[a]
[b]
[b]
Entry
Aryl bromide
DG#
DG#
DDG#
DDG#
a
b
calcd
exp
1
2
3
4
5
6
7
8
9
1a
1b
1c
1d
1e
1h
1i
1l
1m
1o
23.5
20.4
20.2
19.5
20.0
19.9
19.6
21.5
21.1
21.1
19.8
20.7
20.3
20.4
20.6
20.6
20.3
20.7
20.7
19.3
ꢀ3.7
0.3
0.1
0.9
0.6
0.7
0.7
ꢀ0.8
ꢀ0.4
ꢀ1.8
<ꢀ2.3
ꢀ0.07
ꢀ0.07
0.9
Computational experiments were also performed for the
dimer with PCy3 as the phosphine ligand (IId-Cy), and the
corresponding monomer IIe-Cy only exhibits limited protec-
0.6
1.0
0.8
ꢀ0.2
ꢀ
tion of the vacant site by one C H bond of a cyclohexyl
0.3
group (H···Pd 2.475 ꢄ). The equilibrium is slightly in favor
of the monomer (DG=ꢀ1.3 kcalmolꢀ1), but the origin of
this stability is essentially entropic as the corresponding
PCM DE value is 18.3 kcalmolꢀ1. With respect to dissocia-
tion, dimer IId-Cy is 18.1 kcalmolꢀ1 more stable than the
dimer with DavePhos (IId) and therefore, the equilibrium
IId-CyQIIe-Cy+IIe-Cy is likely to have a stronger impact
on the overall reactivity than the corresponding equilibrium
for the DavePhos system. This result might explain why a
higher temperature is required with PCy3 to observe signifi-
cant conversion. The PCy3 ligand does not afford enough
protection of the vacant site on Pd during some crucial steps
and any molecule that is able to interact with the metal
could deactivate the catalyst at room temperature. This
10
ꢀ1.8
[a] Calculated values. [b] DDG# =DG#bꢀDG#
. A negative value for
a
DDG# indicates that the b-arylated product is kinetically preferred.
The trend observed experimentally for the competition
between a- and b-arylation is particularly well reproduced
by the calculations (jDDG#calcdꢀDDG#exp j ꢂ0.7 kcalmolꢀ1).
The results in Table 2 indicate that the a-arylation pathway
is more sensitive to variation in the nature of the aryl ring.
This result is not surprising, as a-arylation involves cleavage
ꢀ
of the Pd Ar bond as the rate-determining step where any
ꢀ
factor that renders the Pd Ar bond stronger would disfavor
the reaction.[31] This influence is indeed what is observed
when electron-withdrawing groups are placed at the ortho
position (Table 1, entries 1, 7, 10), where it has been shown
result explains why [Pd
is not a competent catalyst, as, even if [Pd
(PCy3)] could be generated,[35] there will be a competition
ACHUTGTNERN(NUG 2-FC6H4)(Br)ACTHUNGTRENNNUG
AHCTUNGTRENNUNG
ꢀ
G
that the resulting Pd CAr bond is significantly stronger than
between the enolate and the free PCy3 group for binding to
the vacant site.
when the substituents are placed at other positions.[36] The
activation barrier for a-arylation in the case of compound
1a (Table 2, entry 1) stands in marked contrast with the
other values (Table 2, entries 2–10). In contrast, apart from
the case of compound 1o (Table 2, entry 10), the activation
barriers for b-arylation span a much-narrower range of
values. This result is not surprising as the electronic influ-
As illustrated in Figure 1, the biarylphosphine family of li-
gands is particularly well-adapted to the b-arylation of ester
enolates, as it provides the best compromise between effi-
cient steric protection and easy generation of a reactive
vacant site through isomerization of the biaryl group. How-
ever, the selectivity between the a- and b-arylation path-
ways strongly depends on the nature of the aryl electrophile
(Table 1). To shed more light on these results, the competi-
tion between the a- and b-arylation pathways was studied
computationally for the reaction of ester 2a with selected
aryl bromides. The only difference between the computed
and experimental systems was in the nature of the ester
group (Me versus tBu), but the latter was already shown ex-
perimentally to have only a minor effect on the reaction.
The a- and b-arylation pathways were studied in detail for
m- (1b) and p-fluorobromobenzene (1c) and the results
were similar to those obtained with o-fluorobromobenzene
(1a), with the highest point on the b-arylation pathway also
ꢀ
ence of the substituent on the cleavage of the Pd Ar bond
occurs after the rate-determining step. As already discussed,
the C C reductive-coupling step along the b-arylation path-
ꢀ
way is less affected by the electronic properties of the aryl
bromide because it benefits from the stabilizing influence of
an a-CH agostic interaction and from the interaction be-
tween the biaryl group of DavePhos and Pd. The results in
Table 2 show that, as soon as no electron-withdrawing group
is present in the ortho position of the aryl bromide, the a-
and b-arylation pathways are associated with activation bar-
riers of similar magnitude. However, introducing electron-
withdrawing groups onto the meta and/or para position(s) of
the aryl ring seems to destabilize the a-arylation route suffi-
1942
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1932 – 1944

Mechanism of the Palladium-Catalyzed b-Arylation of Ester Enolates
FULL PAPER
ciently to increase the proportion of b-arylated product to a
significant extent (Table 2, entries 8–10).
The trend in selectivity observed experimentally with differ-
ently substituted aryl bromides agreed well with that ob-
served from the calculations. The presence of electron-with-
drawing groups on these bromides mainly affected the a-ar-
Biaryldialkylphosphines seem to constitute the most-well-
adapted family of ligands for achieving significant selectivity
in favor of b-arylation at mild temperatures with aryl bro-
mide 1a (Figure 1).[6] For instance, PCy3 requires a higher
temperature and PtBu3 gives mostly a-arylation products.
However, the substitution pattern of the biaryl group has a
critical influence on a/b-arylation selectivity in the case of
aryl bromide 1b (Figure 2), with DavePhos being currently
the optimal ligand. Further studies are needed to better un-
derstand this behavior and to design more selective ligands
for a broader range of b-arylation reactions. In addition, a
detailed analysis of the effect of the enolate structure on the
selectivity of the arylation reaction will be reported in due
course.
ꢀ
ylation pathway by disfavoring C C reductive elimination.
Finally, we concluded that the higher activity of the biaryl-
dialkylphosphine ligands, compared to their corresponding
trialkylphosphines, could be attributed to stabilizing interac-
tions between the biaryl backbone of the ligands and the
metal center, thereby preventing deactivation of the b-aryla-
tion pathway. However, the exact role of the biaryl substitu-
ents, culminating with the unique trend in selectivity ob-
served with DavePhos, is still elusive and will constitute the
matter of future studies.
Acknowledgements
Conclusion
This work was supported by the Universitꢀ Claude Bernard Lyon 1,
ANR (programme blanc “AlCaCHA”) and the Institut Universitaire de
France. We also thank Dr. E. Jeanneau (UCBL) for crystallographic data
collection, structure solution, and refinement.
The palladium-catalyzed b-arylation of ester enolates with
aryl bromides was studied in detail both experimentally and
computationally. First, the effect of the ligand on the selec-
tivity of the a/b-arylation reactions of ortho- and meta-fluo-
robromobenzene (1a and 1b, respectively) was described.
Whereas b-arylation was predominantly observed with o-flu-
orobromobenzene (1a) for a range of biarylphosphine li-
gands, a-arylation was primarily observed with m-fluorobro-
mobenzene (1b) for all ligands except with DavePhos (L2),
which gave an approximate 1:1 mixture of a-/b-arylated
products. Next, the effect of the substitution pattern of the
aryl-bromide reactant was studied with DavePhos as the
ligand. We showed that electronic factors played a major
role in the a/b-arylation selectivity, with electron-withdraw-
ing substituents favoring b-arylation. In addition, electron-
withdrawing substituents had the most influence when they
were at the ortho position, and steric factors contributed
only marginally. The reaction mechanism was studied exper-
imentally with DavePhos as the ligand. Kinetic and deuteri-
um-labeling experiments suggested that the rate-limiting
step of b-arylation was the palladium-enolate-to-homoeno-
[1] a) D. A. Culkin, J. F. Hartwig, Acc. Chem. Res. 2003, 36, 234–245;
b) C. C. C. Johansson, T. J. Colacot, Angew. Chem. 2010, 122, 686–
718; Angew. Chem. Int. Ed. 2010, 49, 676–707; c) F. Bellina, R.
Rossi, Chem. Rev. 2010, 110, 1082–1146.
[2] a) M. Kawatsura, J. F. Hartwig, J. Am. Chem. Soc. 1999, 121, 1473–
1478; b) D. A. Culkin, J. F. Hartwig, J. Am. Chem. Soc. 2001, 123,
5816–5817; c) S. Lee, J. F. Hartwig, J. Org. Chem. 2001, 66, 3402–
3415; d) D. A. Culkin, J. F. Hartwig, J. Am. Chem. Soc. 2002, 124,
9330–9331; e) J. P. Wolkowski, J. F. Hartwig, Angew. Chem. 2002,
114, 4465–4467; Angew. Chem. Int. Ed. 2002, 41, 4289–4291.
[3] a) R. Jazzar, J. Hitce, A. Renaudat, J. Sofack-Kreutzer, O. Baudoin,
Chem. Eur. J. 2010, 16, 2654–2672; b) O. Baudoin, Chem. Soc. Rev.
2011, 40, 4902–4911.
[4] a) D. Shabashov, O. Daugulis, Org. Lett. 2005, 7, 3657–3659;
b) V. G. Zaitsev, D. Shabashov, O. Daugulis, J. Am. Chem. Soc. 2005,
127, 13154–13155; c) B. V. S. Reddy, L. R. Reddy, E. J. Corey, Org.
Lett. 2006, 8, 3391–3394; d) R. Giri, N. Maugel, J.-J. Li, D.-H. Wang,
S. P. Breazzano, L. B. Saunders, J.-Q. Yu, J. Am. Chem. Soc. 2007,
129, 3510–3511; e) M. Wasa, K. M. Engle, J.-Q. Yu, J. Am. Chem.
Soc. 2009, 131, 9886–9887; f) M. Wasa, J.-Q. Yu, Tetrahedron 2010,
66, 4811–4815; g) D. Shabashov, O. Daugulis, J. Am. Chem. Soc.
2010, 132, 3965–3972; h) Y. Feng, G. Chen, Angew. Chem. 2010,
122, 9943; Angew. Chem. Int. Ed. 2010, 49, 958–961; i) Y. Feng, Y.
Wang, B. Landgraf, S. Liu, G. Chen, Org. Lett. 2010, 12, 3414–3417.
[5] M. Jørgensen, S. Lee, X. Liu, J. P. Wolkowski, J. F. Hartwig, J. Am.
Chem. Soc. 2002, 124, 12557–12565.
[6] A. Renaudat, L. Jean-Gꢀrard, R. Jazzar, C. E. Kefalidis, E. Clot, O.
Baudoin, Angew. Chem. 2010, 122, 7419–7423; Angew. Chem. Int.
Ed. 2010, 49, 7261–7265.
[7] For recent reviews: a) D. S. Surry, S. L. Buchwald, Angew. Chem.
2008, 120, 6438–6461; Angew. Chem. Int. Ed. 2008, 47, 6338–6361;
b) D. S. Surry, S. L. Buchwald, Chem. Sci. 2011, 2, 27–50.
[8] H. Tomori, J. M. Fox, S. L. Buchwald, J. Org. Chem. 2000, 65, 5334–
5341.
[9] C. A. Parrish, S. L. Buchwald, J. Org. Chem. 2001, 66, 3820–3827.
[10] C. Han, S. L. Buchwald, J. Am. Chem. Soc. 2009, 131, 7532–7533.
[11] Chieffi, K. Kamikawa, J. ꢄhman, J. M. Fox, S. L. Buchwald, Org.
Lett. 2001, 3, 1897–1900.
[12] J. Yin, M. P. Rainka, X.-X. Zhang, S. L. Buchwald, J. Am. Chem.
Soc. 2002, 124, 1162–1163.
ꢀ
late isomerization (III!V), which occurs by a b H-elimina-
tion, olefin-rotation, and olefin-insertion sequence. A dimer-
ic oxidative-addition complex (IIb) was isolated and struc-
turally characterized. This complex was shown to be not
only catalytically competent, but also more active than the
[Pd2ACHTUNGTRENNUNG(dba)3]/DavePhos in situ mixture. A common mecha-
nism for a- and b-arylation was described by DFT calcula-
tions performed at the B3PW91 level. These calculations
agreed with experimental observations. With DavePhos as
the ligand, the pathway leading to b-arylation was kinetical-
ly favored over the pathway leading to a-arylation, with the
palladium-enolate-to-homoenolate isomerization being the
rate-limiting step of the b-arylation pathway and the TS for
olefin insertion (TS-IVb-V) its highest point. The nature of
the rate-limiting step changed with PCy3 and PtBu3 ligands,
and with the latter, a-arylation became kinetically favored.
Chem. Eur. J. 2012, 18, 1932 – 1944
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1943

E. Clot, O. Baudoin et al.
[13] For comparison, the Tolman cone angles (q) of related PR3 phos-
phines are as follows:[14] PtBu3 1828, PCy3 1708, PiPr3 1608,
PPh3 1438.
[14] a) C. A. Tolman, Chem. Rev. 1977, 77, 313–348; b) T. L. Brown,
K. J. Lee, Coord. Chem. Rev. 1993, 128, 89–116.
[15] Reactions of aryl bromides with an ortho substituent were slower
and thus they were all conducted at 508C instead of 308C.
[16] U. Christmann, R. Vilar, Angew. Chem. 2005, 117, 370–378; Angew.
Chem. Int. Ed. 2005, 44, 366–374.
[17] a) T. E. Barder, S. D. Walker, J. R. Martinelli, S. L. Buchwald, J. Am.
Chem. Soc. 2005, 127, 4685–4696; b) T. E. Barder, M. R. Biscoe,
S. L. Buchwald, Organometallics 2007, 26, 2183–2192.
corresponding products were easier to analyze by NMR spectrosco-
py.
[26] M. Gꢅmez-Gallego, M. A. Sierra, Chem. Rev. 2011, 111, 4857–4963.
[27] For related isomerizations, see: a) J. P. Wolfe, Eur. J. Org. Chem.
2007, 571–582; b) S. Seel, T. Thaler, K. Takatsu, C. Zhang, H. Zipse,
B. F. Straub, P. Mayer, P. Knochel, J. Am. Chem. Soc. 2011, 133,
4774–4777.
[28] A similar explanation was put forward for Pd- and Ni-catalyzed eth-
ylene polymerization reactions: a) L. K. Johnson, C. M. Killian, M.
Brookhart, J. Am. Chem. Soc. 1995, 117, 6414–6415; b) L. K. John-
son, S. Mecking, M. Brookhart, J. Am. Chem. Soc. 1996, 118, 267–
268; c) L. Deng, P. Margl, T. Ziegler, J. Am. Chem. Soc. 1997, 119,
1094–1100; d) L. H. Shultz, M. Brookhart, Organometallics 2001,
20, 3975–3982.
[18] Y. Pan, G. B. Young, J. Organomet. Chem. 1999, 577, 257–264.
[19] For details, see the Supporting Information.
3
ꢀ
[29] For related studies in the context of C
G
[20] With SPhos L4, see: a) M. R. Biscoe, T. E. Barder, S. L. Buchwald,
Angew. Chem. 2007, 119, 7370–7373; Angew. Chem. Int. Ed. 2007,
46, 7232–7235; with BrettPhos L9: b) B. P. Fors, D. A. Watson,
M. R. Biscoe, S. L. Buchwald, J. Am. Chem. Soc. 2008, 130, 13552–
13554; with KenPhos L10: c) X. Shen, G. O. Jones, D. A. Watson, B.
Bhayana, S. L. Buchwald, J. Am. Chem. Soc. 2010, 132, 11278–
11287.
tion, see: a) M. Chaumontet, R. Piccardi, N. Audic, J. Hitce, J.-L. Pe-
glion, E. Clot, O. Baudoin, J. Am. Chem. Soc. 2008, 130, 15157–
15166; b) S. Rousseaux, M. Davi, J. Sofack-Kreutzer, C. Pierre, C. E.
Kefalidis, E. Clot, K. Fagnou, O. Baudoin, J. Am. Chem. Soc. 2010,
132, 10706–10716.
[30] T. E. Barder, S. L. Buchwald, J. Am. Chem. Soc. 2007, 129, 12003–
12010.
[21] This distance corresponds to the sum of the van der Waals radii of
N and Pd atoms (3.18 ꢄ); thus, this interaction should be essentially
electrostatic. For recent examples of square-pyramidal Pd complexes
with an apical nitrogen ligand, see: a) M. Kujime, S. Hikichi, M.
Akita, Chem. Lett. 2003, 32, 486–487; b) X. Zhang, Q. Xia, W.
Chen, Dalton Trans. 2009, 7045–7054 and the references therein.
[22] CCDC-845801 (IIb) and CCDC-845802 (IIc) contain the supplemen-
tary crystallographic data for this paper. These data can be obtained
free of charge from The Cambridge Crystallographic Data Centre
via www.ccdc.cam.ac.uk/data_request/cif.
[31] J. F. Hartwig, Inorg. Chem. 2007, 46, 1936–1947.
[32] The transition state TS-IIIa-IVa-tBu could not be located and it was
assumed to have the same energy as the intermediate IVa-tBu.
[33] G. C. Fu, Acc. Chem. Res. 2008, 41, 1555–1564.
[34] C. E. Kefalidis, O. Baudoin, E. Clot, Dalton Trans. 2010, 39, 10528–
10535.
[35] [Pd
mized and the phosphine dissociation was associated to a PCM
value of DE=15.4 kcalmolꢀ1
ACHTGNUNRETN(UNG 2-FC6H4)(Br)ACHTUNTGRENNUG(PCy3)2] and [PdACHTUNEGRTG(NNUN 2-FC6H4)(Br)ACHTNUGRTEN(NNGU PCy3)] were opti-
.
[36] a) E. Clot, C. Mꢀgret, O. Eisenstein, R. N. Perutz, J. Am. Chem.
Soc. 2009, 131, 7817–7827; b) M. E. Evans, C. L. Burke, S. Yai-
buathes, E. Clot, O. Eisenstein, W. D. Jones, J. Am. Chem. Soc.
2009, 131, 13464–13473; c) T. Tanabe, W. W. Brennessel, E. Clot, O.
Eisenstein, W. D. Jones, Dalton Trans. 2010, 39, 10495–10509; d) J.
Guihaumꢀ, E. Clot, O. Eisenstein, R. N. Perutz, Dalton Trans. 2010,
39, 10510–10519.
[23] a) C. Amatore, A. Jutand, F. Khalil, M. A. M’Barki, L. Mottier, Or-
ganometallics 1993, 12, 3168–3178; for a review: b) I. J. S. Fairlamb,
Org. Biomol. Chem. 2008, 6, 3645–3656.
ꢀ
[24] For a mechanistic study of b H elimination from Pt–enolate com-
plexes, see: E. J. Alexanian, J. F. Hartwig, J. Am. Chem. Soc. 2008,
130, 15627–15635.
[25] Benzyl esters were chosen because they were easier to prepare, and
aryl bromide 1p was chosen instead of compound 1a because the
Received: October 6, 2011
Published online: January 13, 2012
1944
www.chemeurj.org
ꢂ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1932 – 1944